Research On Mobile Phone Customer Segmentation Based On Data Mining

With the rapid development of communication technology,China's three major telecom operators face the huge challenges and opportunities.How to change the existing business operating model,build their advantage and strive for more users can be seen as the inevitable problems that operators should deal with.In order to solve the above problems,they have to know more information about the mobilephone users' current preferences and relevant behaviors.The user segment plan must be conducted to master the mobile-phone users information well.The usersegmenting can be operated through data techniques,which aims to get more valuable customer information and behavior characteristics,grasp the new requirements of customers,make targeted marketing strategy and finally consolidate old customers,and develop new customers.The maximize profits can be achieved by carrying on the fine service by differentiation stratify for different users,improving customer value,and controlling costs.In this paper,the k-means algorithm is studied,and the algorithm is improved by optimizing the initial cluster center,combined with the characteristics of telecom industry and mobile phone users.Researchers establish the customer segmentation model by choosing the mobile phone users' behavior attributes and value attributes as the segmentation variables and analyzing the 4000 sample data,preprocessing the data source of data,removing the outliers,and making data conversion.Secondly,with K-meams clustering algorithm of customer segments,with different clustering is worth to several groups of different clustering results,and then the discriminant analysis method of the clustering results are identified and verified,to determine the sample accuracy are analyzed and compared,so as to obtain the optimal clustering number;Finally,respectively,with the traditional K-means algorithm and the improved k-means algorithm to subdivide the customers,and through analysis and comparison,it is concluded that the final clustering results,and to the ultimate customer classification results for feature analysis and summary,according to the customs and characteristics of different customer groups,corresponding to the marketing strategy,In this paper,the improved algorithm K-meams establish a customer segmentation model,subdivision of customers can be realized.Segmentation results with reasonable and practical,the marketing strategy formulation has a certain practical significance.